Time-domain (TD) pulse laser is employed as the measurement laser to solve the inverse problem of radiative heat transfer and phase change in the participating medium. The finite volume method and an improved stochastic particle swarm optimization (ISPSO) algorithm are employed as the direct and inverse problem algorithms, respectively. Moreover, an optimal selection principle of the TD dimensionless boundary temperature measurement signals based on the principle component analysis (PCA) approach is proposed to improve the retrieval accuracy in retrieving the Stefan number St and the conduction to radiation parameter N. Results show that the TD dimensionless boundary temperature measurement signals obtained within dimensionless time selected within [0.05tp*, 0.8tp*] helps with the improvement in the retrieval accuracy. Moreover, compared with the SPSO algorithm, the ISPSO can avoid local optima and improve convergence accuracy, and reasonable results can be obtained even with 5% random measurement errors. As a whole, the present methodology provides a reliable and effective technique to study the inverse problem of radiative heat transfer and phase change in participating medium.